EU-FP7 QA4ECV: a 35 year ECV of albedos, fapar & LAI and their uncertainties

Transcription

1 EU-FP7 QA4ECV: a 35 year ECV of albedos, fapar & LAI and their uncertainties Folkert Boersma KNMI (Co-ordinator) Jan-Peter Muller UCL (Co-ordinator, Land) Page 1 Version: WP4 Harmonised ECV retrievals & records GlobAlbedo UCM-2 17 QA4ECV Kick-off GlobAlbedo meeting, UCM February th September 14 1 th September , De 1Bilt

2 Page 2 Version: j.muller@ucl.ac.uk For land: data providers often use different instruments and different algorithms to produce the same parameter Case: Albedo for a single location N.B. All EO albedos agree with each other BUT disagree with tower measurements. This is unusual but not uncommon GlobAlbedo UCM-2 17 th September 14 2

3 For Atmosphere Data providers use same instrument but different algorithms Case: NO 2 Huan Yu, BIRA Page 3 Version: GlobAlbedo UCM-2 17 th September 14 3

4 Introduction and motivation User perspective: This is where QA4ECV comes in Page 4 Version: GlobAlbedo UCM-2 17 th September 14 4

5 Mission statement QA4ECV QA4ECV will show how trustable assessments of satellite data quality can facilitate users in judging fitness-forpurpose of the ECV Climate Data Record. rg QA4ECV will provide quality assured long-term Climate Data Records of several ECVs relevant for policy and climate change assessments :56 ESA CCI Aerosol Cloud CMUG Fire GHG Glaciers Ice Sheets Land Cover Ocean Colour Ozone Sea Ice Sea Level SST CCI tions GCOS Calendar Essential Site map Climate Variables The 50 GCOS Essential Climate Variables (ECVs) (2010) are required to support the work of the UNFCCC and the IPCC. All ECVs are technically and economically feasible for systematic observation. It is these variables for which international exchange is required for both current and historical observations. Additional variables required for research purposes are not included in this table. It is emphasized that the ordering within the table is simply for convenience and is not an indicator of Over relative 30 priority. years of global soil moisture observations for climate Domain Atmospheric (over land, sea GCOS Essential Climate Variables Wed, :31 The CCI Soil Moisture project, together with ESA's WACMOS project, has publically released an Essential Climate Variable soil moisture Surface:[1] Air temperature, Wind speed and direction, Water dataset that has been put together by the two projects. The dataset of changes in global soil moisture was generated using active and vapour, Pressure, Precipitation, Surface radiation budget. passive microwave spaceborne-instruments and covers the 32 year period from 1978 to Upper-air:[2] ESA Climate Change Initiative Temperature, Wind speed Read and direction, more» Water vapour, Cloud properties, Earth radiation budget Wed, :03 Climate change is arguably the greatest challenge facing mankind in the twenty-first century. Its importance has been recognised in recent reports from the IPCC and from UNFCCC, and the overwhelming economic consequences are set out in the Stern Report. Read more»

6 Why QA4ECV is necessary Users need clear info on validity of EO/climate data sets Unique records available, but need info on strength/weakness web portal Maturity Software readiness 1 Under development 2 Minor changes Validation 3 Continuous crosschecking Quality Assurance System Provides traceable quality info on EO/climate data Tied to international standards QA tools to support user community in tracing quality Multi-decadal records for atmosphere/terrestrial ECVs There is a need for quality-assured long-term climate data records

7 Detailed objectives of QA4ECV 1. Rigorous QA methodologies for satellite ECV products QA framework applicable to many ECVs SW tools for do-it-yourself QA SI standards as in QA4EO (through NPL) 2. Multi-decadal satellite-derived global ECV records 3 Terrestrial and 3 Atmospheric ECVs w/ global coverage Not yet covered by ESA or EUMETSAT activities; yr 3. Traceable QA applied to ECV retrievals and products QA4ECV approach applied to independent reference data, ECV retrievals, and final products 4. Information on quality and fit-for-purpose nature of datasets QA Office to audit ECV records against GCOS, WMO crit. Assess impact of ECV records for applications

8 How will QA4ECV reach its objectives? Step 1: Develop a Quality Assurance System Users obtain a one-stop-shop for all QA info The QA4ECV Office provides independent verification of ECV QA Traceability to SI & community reference standards through metrology institute Support the retrieval groups with QA tools, software and uncertainty analysis

9 How will QA4ECV reach its objectives? Step 2: Generate multi-decadal ECV records Land ECVs Spectral albedo, LAI & FAPAR Surface and vegetation state Indicators for land use change, biosphere activity Measured since 1980s Evaluate carbon cycle & water cycle in climate models Atmosphere ECVs NO 2, HCHO & CO Air pollutants Drive ozone and aerosol formation Measured since 1995/2000 Evaluate atm. chemistry modules in climate models Provide info on effectiveness of policies, trends in fires etc. ECV sets do exist to some extent, but they are not necessarily coherent and not quality assured.

10 Castellanos and Boersma, Sci. Rep., 2012 Policy control with OMI NO 2 measurements / /2005 Continuous reduction ( ): -2% to -5% Recession Change ( ): -15% to -30% 2008/ / /2005 Page 10 Version: GlobAlbedo UCM-2 17 th September 14 10

11 How will QA4ECV reach its objectives? Step 3: Apply the QA System to ECVs QA of independent reference data themselves QA of the ECV retrieval algorithm (steps) QA of complete ECV record & test against GCOS criteria

12 How will QA4ECV reach its objectives? Step 4: Interact with the users From the beginning by involving users and suppliers in the design of the QA4ECV Office / workshops Throughout the project by interaction with users on the functionalities ( do it yourself QA) and its applicability across (other) ECVs To the end QA reports on on the ECV data records (green) QA reports on retrieval algorithms (yellow)

13 Contribution of QA4ECV to GMES Climate Service FP7-SPACE : contribute toward the (pre-) operational capacities of GMES, by augmenting the number of currently available quality-assured long term ECV records and by providing methodologies suitable for reliable assessments of the climate quality of ECV products. A clear tested blueprint for an implementable pre-operational Quality Assurance (QA) service to underpin future European-wide multi-ecv Climate Services. A fully automated validation system enabling independent stakeholder assessment of QA of Atmosphere ECV products from different producers. A set of tools for the production, auditing and use of quality information for ECV records

14 The ECVs to be generated 1. Spectral albedo Radiation budgets, AOT, LAI, FAPAR 2. LAI Radiation budgets, CO 2, soil moisture 3. FAPAR Radiation budgets, CO 2, soil moisture 2010 Implementation Plan for GCOS: parties are called to develop and implement coordinated and complementary strategies for long-term measurements of ozone and precursor species (NO 2, SO 2, HCHO, and CO) Development of long-term datasets based on the currently available measurements from the past 15 to 20 years (GCOS- 154, section , page 42). 4. NO 2 O 3 (T+S), AOT, AAI, albedo 5. HCHO O 3 (T), AOT, AAI, surface T, albedo, LAI 6. CO O 3 (T), AOT, AAI, surface T

15 QA4ECV Aims Global Land surface products Produce 35 years ( ) of Land Surface Spectral & BroadBand BRDF/Albedo, fapar, LAI every day at 1, 5, 25km & monthly 0.05º & 0.5º from European & US space assets to generate Essential Climate Variables. Quality Assurance at each pixel using optimal estimation. Input data level 1b (radiometrically calibrated, satellite projection) with a priori climatology derived from MODIS C5 BRDF (3/2000-3/2014) GEO (SCOPE-CM : ) & AVHRR (LTDR : ) VEGETATION (3/ /2013) & PROBA-V (9/ /2016) MERIS (6/2002-4/12) & ATSR2 6/95-3/00, AATSR 6/02-4/12 Level 2 only: MODIS C6 (3/ /2016), MISR (3/ /2016) Validation of final albedo products & intermediate products (e.g. sensor-to-sensor, cloud masks, aerosol AOD, narrow-tobroadband) 726Tb of CEMS space (NERC Big Data) - all products to be freely available via wget/curl, http, WPS and display browse & animations Page 15 Version: GlobAlbedo UCM-2 17 th September 14 15

16 Prior Art - Example LAI derived from GlobAlbedo using TIP* Completed fapar & LAI for inc. uncertainties * Pinty et al. (2007) J. Geophys. Res.-Atmos. 112 (D10), D10116 Page 16 Version: GlobAlbedo UCM-2 17 th September 14 16

17 Page 17 Version: Intercomparison of LAI derived from GlobAlbedo vs MODIS GlobAlbedo MODIS GlobAlbedo UCM-2 17 th September 14 17

18 Re-processed MODIS Prior using all spectral and 3 broadbands Processed on CEMS in 3 days(81 times faster than on 8-core blade) Completed all 326 tiles at 500m for 14 years of MODIS data Once Collection 6 daily products available, will re-process to obtain daily prior Page 18 Version: GlobAlbedo UCM-2 17 th September 14 18

19 Page 19 Version: Global MODIS Prior in False Colour GlobAlbedo UCM-2 17 th September 14 19

20 Example uses of CEMS-JASMIN for global land surface products Objective 1: Re-project BRDF files from SIN-coordinates to lat/lon using an Energy Conservation method Challenge: the projected SIN-Tiles into lat/lon results for non-rectangular shapes, with different SIN tiles Solution: SIN and Lat,Lon Cells are represented by geometry polygons rather than simple points and then the process is based on ratios of common area rather than on simple distance Challenge: huge number of polygons to be spatiality indexed and processed. This process requires massive RAM and usually takes a very long time! Solution: Use Cloud-computing system on CEMS-JASMIN (~100 times faster than 224-core in house linux cluster!) Objective 2: Create specific albedo products for computation of 8-daily LAI/fAPAR between 2002 and 2011 at 3 different resolutions: 1km, 5km and 25km Challenge: Upscale big data BRDF (50TB) from 1km to 5km and 25km using energy conservation method, and then create separate Albedo-Snow_only and Albedo-Snow_Free products: This process is extremely time consuming! Solution: Cloud-computing system in CEMS-JASMIN (~100 times faster than 224-core in house linux cluster) Page 20 Version: GlobAlbedo UCM-2 17 th September 14 20

21 CEMS-JASMIN MSSL-UCL Example use of CEMS-JASMIN(2) for near-line subsetting Query - Region of interest - Time period - Product/Bands - Landcover classes - address On-line statistics (Albedo, LAI/fAPAR, filtered by landcover) by Region of Interest: point, line or polygon. URLs by Rewrite and stock Queries (first come, first serve) Reproject Region of interest from lat,lon to SIN Extract, Filter, Calculate Statistics csv files & Plots (timeseries) Albedo fapar LAI Page 21 Version: Being integrated into Microsoft-Azure GlobAlbedo UCM-2 17 th September 14 21

22 Where are we now and where do we go next? Page 22 Version: All GlobAlbedo BBDRs and BRDFs copied onto CEMS after 2nd re-processing All MODIS Collection 5 (MOD09 & MYD09) BRFs downloaded (25 Mbytes/day) to CEMS. Will download Collection 6 when available (from Oct 2014) All MISR (MISR2AS) BRFs downloaded. All AVHRR LTDRs available MERIS & ATSR2 & AATSR level-1b already available. VEGETATION level-1b about to be copied onto CEMS GEO data from METEOESAT is being processed for a test year into BRFs. Testing software for ingestion into data fusion GEO data from GOES & GMS is being processed into BRFs FastOpt processed 10 years of GlobAlbedo data into fapar & LAI All previous products available through GlobAlbedo.org website Systematic processing of 35 years starts in July 2015 Monte Carlo ray-tracing being employed to derive simulated validation test datasets GlobAlbedo UCM-2 17 th September 14 22

23 Anticipated users of QA4ECV products Data users from Land community - Researchers - Authorities (forest management, urban planning) - Policy makers (assessment of trends, history) Data users from Atmosphere community - Researchers - Authorities (Public health, climate) - Policy makers (trends, chemistry-climate feedbacks) QA system users (ECV developers and users) WP1 provides formal links with other ECV projects, within the EUMETSAT SAF Network and ESA CCI to ensure that the system developed in QA4ECV has wider applicability.

24 Anticipated users of QA4ECV products Our own project Interface between albedo and atmosphere retrievals Climate modellers to assess fitness-for-purpose of Land ECVs for climate research and modelling Chemistry-climate modellers to assess fitness-for-purpose of Atmosphere ECVs for AQ, chemistry-climate, and trends MACC-II & GIO Global Land S5P and new geostationary sensors (TEMPO, S4, GEMS)

25 QA4ECV users Obtain data through: Project website - Needs/requirements have been identified from the start of the project; user survey, set-up helpdesk QA4ECV obtains user needs through Maintain active dialogue with users and key projects to ensure dissemination and use of QA system Coordinated outreach with other 4 other FP7 Space projects Feedback will be organized through Helpdesk Links to relevant projects (ESA CCI, EUMETSAT SAF) Within our project (fitness-for-purpose WP)

26 Expected impacts 1. Concept service to underpin European-wide ECV services 2. Validation system (QA independent ref., retrieval & product) 3. Tools for do-it-yourself QA 4. Quality assured multi-decadal records Land & Atmosphere 1. Interaction with policy makers and users at start and throughout the project 2. Ingestion of Land ECV records in (climate) models will add to European Clearing House Mechanism